[ Quoting Denis Tonn to Scott Little ]
 SL> 13.5 meg, with around 950k free. Win will try to keep this memory free 
o
 DT> that SL> it is instantly available. When something starts using LOTs of 
memory
 DT> Better. It's not a "simple" algorithm. I am curious how the algorithm
 DT> takes into consideration "dormant" application pages vs "in use" cache
 DT> pages. Judging by the numbers you have supplied above, I would guess
 DT> that it doesn't look at the system working set, but shrinks the cache
 DT> size down to the minumum before it would start swapping application
 DT> or system pages. Not the most effective use of the RAM.
It might be based on time. New DOS boxes usually pop up very quickly on my
computer (via hotkeys). But if I leave the system on overnight, with no 
windows
open, it takes a few seconds for it to open a new one. Once I have one window
in memory, everything is as fast as usual.
 DT> Have you done any tests with various sizes? Min/max both at 2MB, 4MB,
 DT> 6MB, etc. Min/Max at 2MB/4MB, 2MB/8MB, 4MB/8MB, etc.. First time
 DT> opening of apps vs multiple opens and speeds of data intensive apps?
I will have a go at it, probably on Tuesday (day after next) as I have a day
off. I am getting a copy of Win95-B (and Merlin) sometime this week, so I'll
repeat the same tests on FAT32 as well.
Does Merlin come with HPFS386? I got 32meg RAM to burn, I might as well have
a decent cache.
Regards,
 - Scott
[ admin@cyberia.asstdc.com.au | www.asstdc.com.au/~cyberia ]
--- FMailX32 1.22
3:712/848)
---------------